*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
* Copyright (c) 2013 Steven Hartland. All rights reserved.
* Copyright (c) 2013 by Joyent, Inc. All rights reserved.
+ * Copyright (c) 2016 Actifio, Inc. All rights reserved.
*/
#include <sys/zfs_context.h>
#include <sys/dsl_userhold.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_synctask.h>
+#include <sys/dsl_destroy.h>
#include <sys/dmu_tx.h>
#include <sys/dsl_pool.h>
#include <sys/dsl_dir.h>
#include <sys/zfs_ioctl.h>
#include <sys/dsl_deleg.h>
#include <sys/dmu_impl.h>
-
-typedef struct dmu_snapshots_destroy_arg {
- nvlist_t *dsda_snaps;
- nvlist_t *dsda_successful_snaps;
- boolean_t dsda_defer;
- nvlist_t *dsda_errlist;
-} dmu_snapshots_destroy_arg_t;
+#include <sys/zvol.h>
+#include <sys/zcp.h>
int
dsl_destroy_snapshot_check_impl(dsl_dataset_t *ds, boolean_t defer)
return (0);
}
-static int
+int
dsl_destroy_snapshot_check(void *arg, dmu_tx_t *tx)
{
- dmu_snapshots_destroy_arg_t *dsda = arg;
+ dsl_destroy_snapshot_arg_t *ddsa = arg;
+ const char *dsname = ddsa->ddsa_name;
+ boolean_t defer = ddsa->ddsa_defer;
+
dsl_pool_t *dp = dmu_tx_pool(tx);
- nvpair_t *pair;
int error = 0;
+ dsl_dataset_t *ds;
- if (!dmu_tx_is_syncing(tx))
- return (0);
-
- for (pair = nvlist_next_nvpair(dsda->dsda_snaps, NULL);
- pair != NULL; pair = nvlist_next_nvpair(dsda->dsda_snaps, pair)) {
- dsl_dataset_t *ds;
-
- error = dsl_dataset_hold(dp, nvpair_name(pair),
- FTAG, &ds);
-
- /*
- * If the snapshot does not exist, silently ignore it
- * (it's "already destroyed").
- */
- if (error == ENOENT)
- continue;
+ error = dsl_dataset_hold(dp, dsname, FTAG, &ds);
- if (error == 0) {
- error = dsl_destroy_snapshot_check_impl(ds,
- dsda->dsda_defer);
- dsl_dataset_rele(ds, FTAG);
- }
+ /*
+ * If the snapshot does not exist, silently ignore it, and
+ * dsl_destroy_snapshot_sync() will be a no-op
+ * (it's "already destroyed").
+ */
+ if (error == ENOENT)
+ return (0);
- if (error == 0) {
- fnvlist_add_boolean(dsda->dsda_successful_snaps,
- nvpair_name(pair));
- } else {
- fnvlist_add_int32(dsda->dsda_errlist,
- nvpair_name(pair), error);
- }
+ if (error == 0) {
+ error = dsl_destroy_snapshot_check_impl(ds, defer);
+ dsl_dataset_rele(ds, FTAG);
}
- pair = nvlist_next_nvpair(dsda->dsda_errlist, NULL);
- if (pair != NULL)
- return (fnvpair_value_int32(pair));
-
- return (0);
+ return (error);
}
struct process_old_arg {
if (clone->ds_dir->dd_origin_txg > mintxg) {
dsl_deadlist_remove_key(&clone->ds_deadlist,
mintxg, tx);
+ if (dsl_dataset_remap_deadlist_exists(clone)) {
+ dsl_deadlist_remove_key(
+ &clone->ds_remap_deadlist, mintxg, tx);
+ }
dsl_dataset_remove_clones_key(clone, mintxg, tx);
}
dsl_dataset_rele(clone, FTAG);
kmem_free(zc, sizeof (zap_cursor_t));
}
+static void
+dsl_destroy_snapshot_handle_remaps(dsl_dataset_t *ds, dsl_dataset_t *ds_next,
+ dmu_tx_t *tx)
+{
+ dsl_pool_t *dp = ds->ds_dir->dd_pool;
+
+ /* Move blocks to be obsoleted to pool's obsolete list. */
+ if (dsl_dataset_remap_deadlist_exists(ds_next)) {
+ if (!bpobj_is_open(&dp->dp_obsolete_bpobj))
+ dsl_pool_create_obsolete_bpobj(dp, tx);
+
+ dsl_deadlist_move_bpobj(&ds_next->ds_remap_deadlist,
+ &dp->dp_obsolete_bpobj,
+ dsl_dataset_phys(ds)->ds_prev_snap_txg, tx);
+ }
+
+ /* Merge our deadlist into next's and free it. */
+ if (dsl_dataset_remap_deadlist_exists(ds)) {
+ uint64_t remap_deadlist_object =
+ dsl_dataset_get_remap_deadlist_object(ds);
+ ASSERT(remap_deadlist_object != 0);
+
+ mutex_enter(&ds_next->ds_remap_deadlist_lock);
+ if (!dsl_dataset_remap_deadlist_exists(ds_next))
+ dsl_dataset_create_remap_deadlist(ds_next, tx);
+ mutex_exit(&ds_next->ds_remap_deadlist_lock);
+
+ dsl_deadlist_merge(&ds_next->ds_remap_deadlist,
+ remap_deadlist_object, tx);
+ dsl_dataset_destroy_remap_deadlist(ds, tx);
+ }
+}
+
void
dsl_destroy_snapshot_sync_impl(dsl_dataset_t *ds, boolean_t defer, dmu_tx_t *tx)
{
-#ifdef ZFS_DEBUG
- int err;
-#endif
- spa_feature_t f;
int after_branch_point = FALSE;
dsl_pool_t *dp = ds->ds_dir->dd_pool;
objset_t *mos = dp->dp_meta_objset;
dsl_dataset_t *ds_prev = NULL;
- uint64_t obj, old_unique, used = 0, comp = 0, uncomp = 0;
- dsl_dataset_t *ds_next, *ds_head, *hds;
-
+ uint64_t obj;
ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock));
+ rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
ASSERT3U(dsl_dataset_phys(ds)->ds_bp.blk_birth, <=, tx->tx_txg);
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
ASSERT(refcount_is_zero(&ds->ds_longholds));
if (defer &&
obj = ds->ds_object;
- for (f = 0; f < SPA_FEATURES; f++) {
+ for (spa_feature_t f = 0; f < SPA_FEATURES; f++) {
if (ds->ds_feature_inuse[f]) {
dsl_dataset_deactivate_feature(obj, f, tx);
ds->ds_feature_inuse[f] = B_FALSE;
}
}
+ dsl_dataset_t *ds_next;
+ uint64_t old_unique;
+ uint64_t used = 0, comp = 0, uncomp = 0;
+
VERIFY0(dsl_dataset_hold_obj(dp,
dsl_dataset_phys(ds)->ds_next_snap_obj, FTAG, &ds_next));
ASSERT3U(dsl_dataset_phys(ds_next)->ds_prev_snap_obj, ==, obj);
dsl_deadlist_merge(&ds_next->ds_deadlist,
dsl_dataset_phys(ds)->ds_deadlist_obj, tx);
}
+
dsl_deadlist_close(&ds->ds_deadlist);
dsl_deadlist_free(mos, dsl_dataset_phys(ds)->ds_deadlist_obj, tx);
dmu_buf_will_dirty(ds->ds_dbuf, tx);
dsl_dataset_phys(ds)->ds_deadlist_obj = 0;
+ dsl_destroy_snapshot_handle_remaps(ds, ds_next, tx);
+
/* Collapse range in clone heads */
dsl_dataset_remove_clones_key(ds,
dsl_dataset_phys(ds)->ds_creation_txg, tx);
ASSERT3P(ds_next->ds_prev, ==, NULL);
/* Collapse range in this head. */
+ dsl_dataset_t *hds;
VERIFY0(dsl_dataset_hold_obj(dp,
dsl_dir_phys(ds->ds_dir)->dd_head_dataset_obj, FTAG, &hds));
dsl_deadlist_remove_key(&hds->ds_deadlist,
dsl_dataset_phys(ds)->ds_creation_txg, tx);
+ if (dsl_dataset_remap_deadlist_exists(hds)) {
+ dsl_deadlist_remove_key(&hds->ds_remap_deadlist,
+ dsl_dataset_phys(ds)->ds_creation_txg, tx);
+ }
dsl_dataset_rele(hds, FTAG);
} else {
}
/* remove from snapshot namespace */
+ dsl_dataset_t *ds_head;
ASSERT(dsl_dataset_phys(ds)->ds_snapnames_zapobj == 0);
VERIFY0(dsl_dataset_hold_obj(dp,
dsl_dir_phys(ds->ds_dir)->dd_head_dataset_obj, FTAG, &ds_head));
#ifdef ZFS_DEBUG
{
uint64_t val;
+ int err;
err = dsl_dataset_snap_lookup(ds_head,
ds->ds_snapname, &val);
dmu_object_free_zapified(mos, obj, tx);
}
-static void
+void
dsl_destroy_snapshot_sync(void *arg, dmu_tx_t *tx)
{
- dmu_snapshots_destroy_arg_t *dsda = arg;
- dsl_pool_t *dp = dmu_tx_pool(tx);
- nvpair_t *pair;
-
- for (pair = nvlist_next_nvpair(dsda->dsda_successful_snaps, NULL);
- pair != NULL;
- pair = nvlist_next_nvpair(dsda->dsda_successful_snaps, pair)) {
- dsl_dataset_t *ds;
+ dsl_destroy_snapshot_arg_t *ddsa = arg;
+ const char *dsname = ddsa->ddsa_name;
+ boolean_t defer = ddsa->ddsa_defer;
- VERIFY0(dsl_dataset_hold(dp, nvpair_name(pair), FTAG, &ds));
+ dsl_pool_t *dp = dmu_tx_pool(tx);
+ dsl_dataset_t *ds;
- dsl_destroy_snapshot_sync_impl(ds, dsda->dsda_defer, tx);
- dsl_dataset_rele(ds, FTAG);
- }
+ int error = dsl_dataset_hold(dp, dsname, FTAG, &ds);
+ if (error == ENOENT)
+ return;
+ ASSERT0(error);
+ dsl_destroy_snapshot_sync_impl(ds, defer, tx);
+ zvol_remove_minors(dp->dp_spa, dsname, B_TRUE);
+ dsl_dataset_rele(ds, FTAG);
}
/*
dsl_destroy_snapshots_nvl(nvlist_t *snaps, boolean_t defer,
nvlist_t *errlist)
{
- dmu_snapshots_destroy_arg_t dsda;
- int error;
- nvpair_t *pair;
-
- pair = nvlist_next_nvpair(snaps, NULL);
- if (pair == NULL)
+ if (nvlist_next_nvpair(snaps, NULL) == NULL)
return (0);
- dsda.dsda_snaps = snaps;
- VERIFY0(nvlist_alloc(&dsda.dsda_successful_snaps,
- NV_UNIQUE_NAME, KM_SLEEP));
- dsda.dsda_defer = defer;
- dsda.dsda_errlist = errlist;
+ /*
+ * lzc_destroy_snaps() is documented to take an nvlist whose
+ * values "don't matter". We need to convert that nvlist to
+ * one that we know can be converted to LUA. We also don't
+ * care about any duplicate entries because the nvlist will
+ * be converted to a LUA table which should take care of this.
+ */
+ nvlist_t *snaps_normalized;
+ VERIFY0(nvlist_alloc(&snaps_normalized, 0, KM_SLEEP));
+ for (nvpair_t *pair = nvlist_next_nvpair(snaps, NULL);
+ pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) {
+ fnvlist_add_boolean_value(snaps_normalized,
+ nvpair_name(pair), B_TRUE);
+ }
- error = dsl_sync_task(nvpair_name(pair),
- dsl_destroy_snapshot_check, dsl_destroy_snapshot_sync,
- &dsda, 0, ZFS_SPACE_CHECK_NONE);
- fnvlist_free(dsda.dsda_successful_snaps);
+ nvlist_t *arg;
+ VERIFY0(nvlist_alloc(&arg, 0, KM_SLEEP));
+ fnvlist_add_nvlist(arg, "snaps", snaps_normalized);
+ fnvlist_free(snaps_normalized);
+ fnvlist_add_boolean_value(arg, "defer", defer);
+
+ nvlist_t *wrapper;
+ VERIFY0(nvlist_alloc(&wrapper, 0, KM_SLEEP));
+ fnvlist_add_nvlist(wrapper, ZCP_ARG_ARGLIST, arg);
+ fnvlist_free(arg);
+
+ const char *program =
+ "arg = ...\n"
+ "snaps = arg['snaps']\n"
+ "defer = arg['defer']\n"
+ "errors = { }\n"
+ "has_errors = false\n"
+ "for snap, v in pairs(snaps) do\n"
+ " errno = zfs.check.destroy{snap, defer=defer}\n"
+ " zfs.debug('snap: ' .. snap .. ' errno: ' .. errno)\n"
+ " if errno == ENOENT then\n"
+ " snaps[snap] = nil\n"
+ " elseif errno ~= 0 then\n"
+ " errors[snap] = errno\n"
+ " has_errors = true\n"
+ " end\n"
+ "end\n"
+ "if has_errors then\n"
+ " return errors\n"
+ "end\n"
+ "for snap, v in pairs(snaps) do\n"
+ " errno = zfs.sync.destroy{snap, defer=defer}\n"
+ " assert(errno == 0)\n"
+ "end\n"
+ "return { }\n";
+
+ nvlist_t *result = fnvlist_alloc();
+ int error = zcp_eval(nvpair_name(nvlist_next_nvpair(snaps, NULL)),
+ program,
+ B_TRUE,
+ 0,
+ zfs_lua_max_memlimit,
+ nvlist_next_nvpair(wrapper, NULL), result);
+ if (error != 0) {
+ char *errorstr = NULL;
+ (void) nvlist_lookup_string(result, ZCP_RET_ERROR, &errorstr);
+ if (errorstr != NULL) {
+ zfs_dbgmsg(errorstr);
+ }
+ return (error);
+ }
+ fnvlist_free(wrapper);
- return (error);
+ /*
+ * lzc_destroy_snaps() is documented to fill the errlist with
+ * int32 values, so we need to covert the int64 values that are
+ * returned from LUA.
+ */
+ int rv = 0;
+ nvlist_t *errlist_raw = fnvlist_lookup_nvlist(result, ZCP_RET_RETURN);
+ for (nvpair_t *pair = nvlist_next_nvpair(errlist_raw, NULL);
+ pair != NULL; pair = nvlist_next_nvpair(errlist_raw, pair)) {
+ int32_t val = (int32_t)fnvpair_value_int64(pair);
+ if (rv == 0)
+ rv = val;
+ fnvlist_add_int32(errlist, nvpair_name(pair), val);
+ }
+ fnvlist_free(result);
+ return (rv);
}
int
struct killarg *ka = arg;
dmu_tx_t *tx = ka->tx;
- if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp))
+ if (bp == NULL || BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp))
return (0);
if (zb->zb_level == ZB_ZIL_LEVEL) {
ka.ds = ds;
ka.tx = tx;
VERIFY0(traverse_dataset(ds,
- dsl_dataset_phys(ds)->ds_prev_snap_txg, TRAVERSE_POST,
- kill_blkptr, &ka));
+ dsl_dataset_phys(ds)->ds_prev_snap_txg, TRAVERSE_POST |
+ TRAVERSE_NO_DECRYPT, kill_blkptr, &ka));
ASSERT(!DS_UNIQUE_IS_ACCURATE(ds) ||
dsl_dataset_phys(ds)->ds_unique_bytes == 0);
}
-typedef struct dsl_destroy_head_arg {
- const char *ddha_name;
-} dsl_destroy_head_arg_t;
-
int
dsl_destroy_head_check_impl(dsl_dataset_t *ds, int expected_holds)
{
return (0);
}
-static int
+int
dsl_destroy_head_check(void *arg, dmu_tx_t *tx)
{
dsl_destroy_head_arg_t *ddha = arg;
for (t = 0; t < DD_USED_NUM; t++)
ASSERT0(dsl_dir_phys(dd)->dd_used_breakdown[t]);
+ if (dd->dd_crypto_obj != 0) {
+ dsl_crypto_key_destroy_sync(dd->dd_crypto_obj, tx);
+ (void) spa_keystore_unload_wkey_impl(dp->dp_spa, dd->dd_object);
+ }
+
VERIFY0(zap_destroy(mos, dsl_dir_phys(dd)->dd_child_dir_zapobj, tx));
VERIFY0(zap_destroy(mos, dsl_dir_phys(dd)->dd_props_zapobj, tx));
VERIFY0(dsl_deleg_destroy(mos, dsl_dir_phys(dd)->dd_deleg_zapobj, tx));
dsl_destroy_head_sync_impl(dsl_dataset_t *ds, dmu_tx_t *tx)
{
dsl_pool_t *dp = dmu_tx_pool(tx);
- spa_feature_t f;
objset_t *mos = dp->dp_meta_objset;
uint64_t obj, ddobj, prevobj = 0;
boolean_t rmorigin;
- objset_t *os;
ASSERT3U(dsl_dataset_phys(ds)->ds_num_children, <=, 1);
ASSERT(ds->ds_prev == NULL ||
dsl_dataset_phys(ds->ds_prev)->ds_next_snap_obj != ds->ds_object);
+ rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
ASSERT3U(dsl_dataset_phys(ds)->ds_bp.blk_birth, <=, tx->tx_txg);
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock));
/* We need to log before removing it from the namespace. */
obj = ds->ds_object;
- for (f = 0; f < SPA_FEATURES; f++) {
+ for (spa_feature_t f = 0; f < SPA_FEATURES; f++) {
if (ds->ds_feature_inuse[f]) {
dsl_dataset_deactivate_feature(obj, f, tx);
ds->ds_feature_inuse[f] = B_FALSE;
/*
* Destroy the deadlist. Unless it's a clone, the
- * deadlist should be empty. (If it's a clone, it's
- * safe to ignore the deadlist contents.)
+ * deadlist should be empty since the dataset has no snapshots.
+ * (If it's a clone, it's safe to ignore the deadlist contents
+ * since they are still referenced by the origin snapshot.)
*/
dsl_deadlist_close(&ds->ds_deadlist);
dsl_deadlist_free(mos, dsl_dataset_phys(ds)->ds_deadlist_obj, tx);
dmu_buf_will_dirty(ds->ds_dbuf, tx);
dsl_dataset_phys(ds)->ds_deadlist_obj = 0;
+ if (dsl_dataset_remap_deadlist_exists(ds))
+ dsl_dataset_destroy_remap_deadlist(ds, tx);
+
+ objset_t *os;
VERIFY0(dmu_objset_from_ds(ds, &os));
if (!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY)) {
ASSERT(!DS_UNIQUE_IS_ACCURATE(ds) ||
dsl_dataset_phys(ds)->ds_unique_bytes == used);
+ rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
bptree_add(mos, dp->dp_bptree_obj,
&dsl_dataset_phys(ds)->ds_bp,
dsl_dataset_phys(ds)->ds_prev_snap_txg,
used, comp, uncomp, tx);
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
dsl_dir_diduse_space(ds->ds_dir, DD_USED_HEAD,
-used, -comp, -uncomp, tx);
dsl_dir_diduse_space(dp->dp_free_dir, DD_USED_HEAD,
}
}
-static void
+void
dsl_destroy_head_sync(void *arg, dmu_tx_t *tx)
{
dsl_destroy_head_arg_t *ddha = arg;
VERIFY0(dsl_dataset_hold(dp, ddha->ddha_name, FTAG, &ds));
dsl_destroy_head_sync_impl(ds, tx);
+ zvol_remove_minors(dp->dp_spa, ddha->ddha_name, B_TRUE);
dsl_dataset_rele(ds, FTAG);
}
* remove the objects from open context so that the txg sync
* is not too long.
*/
- error = dmu_objset_own(name, DMU_OST_ANY, B_FALSE, FTAG, &os);
+ error = dmu_objset_own(name, DMU_OST_ANY, B_FALSE, B_FALSE,
+ FTAG, &os);
if (error == 0) {
- uint64_t obj;
uint64_t prev_snap_txg =
dsl_dataset_phys(dmu_objset_ds(os))->
ds_prev_snap_txg;
- for (obj = 0; error == 0;
+ for (uint64_t obj = 0; error == 0;
error = dmu_object_next(os, &obj, FALSE,
prev_snap_txg))
(void) dmu_free_long_object(os, obj);
/* sync out all frees */
txg_wait_synced(dmu_objset_pool(os), 0);
- dmu_objset_disown(os, FTAG);
+ dmu_objset_disown(os, B_FALSE, FTAG);
}
}
objset_t *os;
if (dmu_objset_hold(dsname, FTAG, &os) == 0) {
- boolean_t inconsistent = DS_IS_INCONSISTENT(dmu_objset_ds(os));
+ boolean_t need_destroy = DS_IS_INCONSISTENT(dmu_objset_ds(os));
+
+ /*
+ * If the dataset is inconsistent because a resumable receive
+ * has failed, then do not destroy it.
+ */
+ if (dsl_dataset_has_resume_receive_state(dmu_objset_ds(os)))
+ need_destroy = B_FALSE;
+
dmu_objset_rele(os, FTAG);
- if (inconsistent)
+ if (need_destroy)
(void) dsl_destroy_head(dsname);
}
return (0);
projects / mirror_zfs.git / blobdiff